In this renewal application we extend previous studies on the metastatic phenotype within the stromal microenvironment by taking into consideration heterogeneity among malignant cells within a tumor, and the resulting biological consequences for long-term tumor maintenance (self-renewal) and metastasis. In preliminary studies we have identified a slowly-cycling sub-population within melanomas that gives rise to a rapidly proliferating progeny and possesses self-renewal capacity. Cells of this sub-population are characterized by expression of the histone 3 K4 demethylase JARID1B, which is a member of the highly conserved jumonji family of chromatin regulators. Knockdown of JARID1B in xenografted melanoma cells leads to a decrease in metastases, suggesting JARID1B-positive cells to be potential 'seeds' for tumor maintenance. Expression of JARID1B is dynamically regulated and does not follow a hierarchical cancer stem cell model as JARID1B-negative cells can become positive, and even single melanoma cells, irrespective of any marker expression, are tumorigenic. Because stable down-regulation of JARID1B expression leads, after several consecutive in vivo passages, to loss of tumorigenicity, we are developing in Specific Aim 1 a dynamic equilibrium model for tumor cells. In this model JARID1B expression is required for tumors to self-renew, but JARID-negative cells can regain JARID1B expression depending on the microenvironmental context. Due to their slowly-cycling phenotype, JARID1B-positive cells are expected to survive exogenous stress, e.g. by an initially hostile microenvironment after implantation or chemotherapy. We address in Specific Aim 2 the 'soil' for malignant cells, which is in part provided by the malignant cells themselves and in part by stromal cells. We are testing the hypothesis that expression of JARID1B at the site of metastasis is dynamically regulated through interactions (i) amongst tumor cells and (ii) tumor cells and matrix. For cell-cell signaling, our focus is on Notch 1, because our preliminary studies suggest that Notch 1 is differentially regulated between the slowly-cycling JARID1B-positive subpopulation and the main population. For cell-matrix interactions, we will focus on the matricellular protein tenascin C, which is highly produced by melanoma cells and is expressed by stromal fibroblasts. We expect from these studies a better understanding of the JARID1B-positive population for metastatic progression.